oa Application of 3D Seismic Multi-Attribute and Neural Network Technique for Reservoir Prediction: A Case Study for the Marrat Formation, Kuwait

The use of 3D seismic attributes for predicting reservoir properties away from the well bore has been<br>used routinely in the industry. Recently a study utilizing multi attribute analysis and neutral network<br>technique applied to one of the Marrat reservoirs in west Kuwait has not only described the reservoir<br>geometry but has also opened up new areas for exploration. Further, the seismic derived porosity<br>volume has been also integrated with the geological model for future well placement.<br>The Middle Marrat limestone reservoir of Jurassic age in the Dharif field is one of the major oil<br>producers in the area. This field discovered in 1988, is an elongated anticline trending NNE-SSW, with<br>a major fault to the west. The reservoir thickness varies from 50-230ft and porosities ranging from 12<br>to 20%. Since a Pilot water injection program is being initiated, a good reservoir description would be<br>essential for planning a successful injection program.<br>The seismic derived porosity volume derived from neural network analysis has been a key in identifying<br>inter-well areas as well as regions away from the wells with good porosity which is consistent with the<br>available geological information. Incorporating the porosity volume as a “soft constraint” to the<br>available geological model has further refined the model and is expected to assist in effective<br>placement of future wells.